Experimental Design as a Tool for Optimizing and Predicting the Nanofiltration Performance by Treating Antibiotic-Containing Wastewater

In recent years, there has been an increase in studies regarding nanofiltration-based processes for removing antibiotics and other pharmaceutical compounds from water and wastewater. In this work, a 2^k factorial design with five control factors (antibiotic molecular weight and concentration, nanofiltration (NF) membrane, feed flow rate, and transmembrane pressure) was employed to optimize the NF performance on the treatment of antibiotic-containing wastewater. The resulting multiple linear regression model was used to predict the antibiotic rejections and permeate fluxes. Additional experiments, using the same membranes and the same antibiotics, but under different conditions of transmembrane pressure, feed flow rate, and antibiotic concentration regarding the 2^k factorial design were carried out to validate the model developed. The model was also evaluated as a tertiary treatment of urban wastewater for removing sulfamethoxazole and norfloxacin. Considering all the conditions investigated, the tightest membrane (NF97) showed higher antibiotics rejection (>97%) and lower permeate fluxes. On the contrary, the loose NF270 membrane presented lower rejections to sulfamethoxazole, the smallest antibiotic, varying from 65% to 97%, and permeate fluxes that were about three-fold higher than the NF97 membrane. The good agreement between predicted and experimental values (R² > 0.97) makes the model developed in the present work a tool to predict the NF performance when treating antibiotic-containing wastewater.